ROHF-MP2 optimization failed

Find answers or ask questions regarding Dalton calculations.
Please upload an output file showing the problem, if applicable.
(It is not necessary to upload input files, they can be found in the output file.)

Post Reply
cobalt
Posts: 5
Joined: 24 Sep 2015, 12:30
First name(s): Alfred
Last name(s): Güthler
Affiliation: Dipl.Chem
Country: Germany

ROHF-MP2 optimization failed

Post by cobalt » 24 Sep 2015, 13:06

Hello DALTON Users,

I want to do CASSCF-calculation on Titaniumboride, but before doing that, I have to recognize the "active space". In oder to get this, I did an ROHF/MP2 optimization in one calculation, as shown in the appendix. However it failed and I can only guess why! It seems, that it has something to with the coordinates, but I am no quite sure!
I think it should be possible, to divide the calculations on two different calculations: one optimization and one energy-MP2-calculation. Nevertheless, it would be helpful to realize, why the calculation failed.

I would be kind, if anyone has a tip, about what the problem is!

greetings
Alfred
Attachments
opt_1ord_redint_TiB2_tripl_MP2_Cs_TiB2_cc-pvtz_gew.out
(43.81 KiB) Downloaded 299 times

hjaaj
Posts: 337
Joined: 27 Jun 2013, 18:44
First name(s): Hans Jørgen
Middle name(s): Aagaard
Last name(s): Jensen
Affiliation: Universith of Southern Denmark
Country: Denmark

Re: ROHF-MP2 optimization failed

Post by hjaaj » 28 Sep 2015, 07:54

Your output shows that it is already the high-spin ROHF which fails. Open-shell convergence can be tricky, especially for transition metal molecules, where the auf-bau principle is not always fulfilled.

I suggest you first try to get the correct orbitals by using the MCSCF code instead of the ROHF code. The MCSCF code can also handle Hartree-Fock wave functions (single determinant), and it is a second-order method with very good convergence properties. That is, try this first, and then restart your geometry optimization with ROHF-DIIS+MP2 reading the (hopefully) converged MOs from the MCSCF code. (IF the aufbau principle is not fulfilled for the final orbitals, you need to level-shift your orbital energies.)

Code: Select all

**DALTON INPUT                                    
.RUN WAVEFUNCTION                               
**WAVE FUNCTIONS                                  
.TITLE                                            
 TiB2, symmetry Cs:, initial MCSCF optimization
.MCSCF
*CONFIGURATION INPUT
.SYMMETRY
 1
.SPIN MUL
 3
.INACTIVE
 6 3
.ELECTRONS
 2
.CAS SPACE
 2 0                               
**END OF INPUT      
PS. You write "UHF" in your title line, note that Dalton cannot do UHF calculations, the calculation you asked for is a high spin ROHF calculations.

hjaaj
Posts: 337
Joined: 27 Jun 2013, 18:44
First name(s): Hans Jørgen
Middle name(s): Aagaard
Last name(s): Jensen
Affiliation: Universith of Southern Denmark
Country: Denmark

Re: ROHF-MP2 optimization failed

Post by hjaaj » 28 Sep 2015, 07:55

PS. By the way, if you really want to do a CASSCF calculation, why do you then ask for a slow geometry optimization (because of numerical gradients) with ROHF-MP2 ?

cobalt
Posts: 5
Joined: 24 Sep 2015, 12:30
First name(s): Alfred
Last name(s): Güthler
Affiliation: Dipl.Chem
Country: Germany

Re: ROHF-MP2 optimization failed

Post by cobalt » 10 Oct 2015, 13:42

Hello DALTON-User,

Thank you for your and sorry for my late reply! Mr. Jensen -jaaj-, I had to carefully consider your answer , because CASSCF is a quite new approach to me, with a lot of questions on my side. However I hope that experience will surely answer some of them.
I wanted to to a CASSCF optimization on TiB2 as a test. In order to this I have to recognize the active space. This is new to me, that it is also possible to use a small active space via a CASSCF-calc. and extend this active space cautiously. I expect from an MP2-calculation the orbitals which correlate to each other , identified by the natural orbitals and the their occupation number! However as you suggest, your approach does the same! My problem is how to interpret the results! Your recommended the input produce a multicharged-cation and I am not sure, if this was your intention?! How then do I recognize the active space?

I post the output of the former mentioned input as an attachment!

Greetings from the more or less sunny Germany

Alfred
Attachments
mcscf-cat_TiB2_tripl_Cs_TiB2_def2-tzvp_gew.out
(63.8 KiB) Downloaded 281 times

taylor
Posts: 545
Joined: 15 Oct 2013, 05:37
First name(s): Peter
Middle name(s): Robert
Last name(s): Taylor
Affiliation: Tianjin University
Country: China

Re: ROHF-MP2 optimization failed

Post by taylor » 10 Oct 2015, 16:13

I suspect the input HJAAJ gave earlier was not intended to correspond exactly to your case but to some triplet. I have some additional suggestions but before that I want to clarify something. Your molecule is identified by the program as having Cs symmetry, but the geometry is not far from having C2v symmetry and I do not really understand why a roughly T-shaped TiB2 would be distorted in this way. Is this perhaps some geometry from a builder or some other source that has not been properly symmetrized? If so, you should probably attend to this first.

As far as a starting CASSCF is concerned, transition-metal systems are always difficult but the left-hand end of the period tends to be somewhat easier than the right-hand end. A naive guess at an active space might be all the valence orbitals and electrons on B2 (which is 6 electrons and 8 orbitals) and the 3d and 4s on Ti, which is another 4 electrons and 6 orbitals. Giving a total of 10 electrons in 14 active orbitals, which is a big, but not huge, calculation. However, jumping in with this as the first attempt is likely to use a lot of computer time, especially because one way or another you will have to specify the symmetry, as well as the spin state, of your system, and at this stage this seems to be unknown? In addition, some of the Ti 3d orbitals are likely to be close to empty, which will impact the convergence of the calculation. The electron affinity of B2 is not known from experiment but is significant according to theoretical calculations, so the ground state is likely to have considerable contribution from Ti+/B2-. I think you will have to consider (in a simple-minded pictorial way) the MOs of Ti/B2 and Ti+/B2- to get an idea what the likely occupation is. It is not completely obvious to me, for example, that the ground state is necessarily expected to be a triplet, since at least for Ti/B2 the fragments are both triplets. But I think some pencil-and-paper work will be necessary unless you want to do a lot of calculations to try to understand what is going on! And please check on the symmetry of the molecule.

Best regards
Pete

cobalt
Posts: 5
Joined: 24 Sep 2015, 12:30
First name(s): Alfred
Last name(s): Güthler
Affiliation: Dipl.Chem
Country: Germany

Re: ROHF-MP2 optimization failed

Post by cobalt » 22 Oct 2015, 12:02

Thank you all and Mr. Taylor for your valuable suggestions !

I first tried a CASSCF (6,14) calculation , so my first guess was maybe not too bad! However it did'nt converge and took to much time and memory. I reduced the active space to CAS (6,4) and got some results, that could not be compared with my former findings with another program.
Concerning the high spin state, I did a simple ROHF calc. (actually the singlet cation first) and found that this state has the least energy, but I am fully aware that this could be wrong! The main reason for my caculation is to find the ground state of this Titaniumdiboride!
I tried to calculate the multicharged cation, because I thought , that could be a method, similar to the "Valence Virtual Orbitals" method, but I think i was wrong and the calculation of these orbitals is somewhat different from the common calc. of MO's, morover CASSCF calcs. produce natural orbitals!

As an attachment, I post an attempt of a CASSCF(2,2) on this molecule.
How I can read in the MOs of a former calculation and if necessary, change the order of the MOs ? I also will change the symmetry of the Molecule, which seems to be C2V and not Cs! My first try was to restart the calc. as described in the manual at 5.4, but it failed!? The SIRIUS.RST file could'nt be found.


Thank you also for your suggestions, beyond the program related topics!

Greetings
Alfred
Attachments
MCSCF_TiB2-tripl_Cs_TiB2_def2-TZVP_gew.out
(71.86 KiB) Downloaded 285 times

Post Reply

Who is online

Users browsing this forum: No registered users and 5 guests